Customer Reviews

Wavelets and Filter Banks

5 star:		(5)
4 star:		(2)
3 star:		(0)
2 star:		(0)
1 star:		(4)

 

 

I am presently full time staff and part-time teacher electronics lab and doing research in image processing. I need a new way or new material so as to use wavelets for my research project. It is one of the most readable texts that I have encountered. I was looking for a very applied book and this seems a good one. There is one thing that I found disappointing. I found that in order to understand some things I had to jump to sections further into the book. Ideas are introduced and are not fully explained until later in the book. This leaves the reader puzzled and sometimes very confused when given the first exposure to a topic. I think the development of ideas could have been more cohesive. All in all, it's not a bad book as a reference

My first reaction to Professor Strang's superficially friendly, "just us folks", "you're gonna love this result" style was positive until I realized I wasn't absorbing anything. Could it be me? Well, yes, it is. I just am not a brilliant co-worker who can grasp Strang's giant leaps of intuition and say, "Wow, Gil, I see what you are saying, that really is neat!" This book is written in bursts of enthusiasm with little emphasis on details (like indices on summation operators) and little continuity. It appears to have a layered structure where Strang breezes through an introduction dropping all sorts of sophisticated concepts along with "don't worry, it will all become clear later". Unfortunately at the next layer Strang does not connect up with previous droppings but simply drops more. I am amazed that Matlab would attach this book to their wavelet toolbox, which by the way has a beautiful user's manual. An alternative to this book, "Wavelet and Wavelet Transforms" by Sidney Burrus and friends, is written in a clear step-by-step manner with attention to details (and with a reasonable step size). Upon reading Burrus's book I could go back to Strang's and figure out some of the things he was so pleased with (other than himself).

At many US universities, math departments offer service courses in math for the engineers, and there is a periodic discussion of the curriculum. One difficulty is that the two sides speak different languages;-- in math, it is linear algebra, calculus...,- and in engineering, signals, high-pass/low-pass filters, downsampling/upsampling, filter bank, polyphase matrix...A wonderful feature of the Strang-Nguyen book is that it speaks both languages. In this way it is refreshing, and it stands out

in a class of its own. It has been tested in courses for engineers, and stood the test. From what I hear, it is equally popular in the two cultures, math and engineering.

I think the book is very good compared to other texts that I have perused on the subject. I am presently teaching myself the material so as to use wavelets for my research project. It is one of the most readable texts that I have encountered. I was looking for a very applied book and this seems a good one. There are some proofs but the text is not flooded with functional analysis theorems. There is one thing that I found disappointing. I found that in order to understand some things I had to jump to sections further into the book. Ideas are introduced and are not fully explained until later in the book. This leaves the reader puzzled and sometimes very confused when given the first exposure to a topic .(It is easy to get the wrong idea when you are given a superficial development). I think the development of ideas could have been more cohesive. Other than that, it is a good book.

I just can't believe these reviews: "overall a good text" - doh! I was unfortunate enough to take a course about wavelets for which the professor used this book as the main text (btw the course was offered by Prof. Truong Nguyen, and yes he is one of the authors). After 3 months of painful !struggle! with this book I was sure about only one thing: I could not learn anything about wavelets. First of all, there is no logical flow, as a matter of fact there is no flow of any kind. The authors just leap forward and backward leaving lots of unanswered questions, which is in my humble opinion, a very annoying thing that should not happen while using a text book. Clear statements about what the hell is going on are very rare. Most of the time you are left with vague statements which tell only part of the story. Right now I am in the third year of my ph.d and I am yet to meet an electrical engineer who thinks this book is actually useful in some sense. Don't waste your money on it. Start with "Wavelet and Wavelet Transforms" Burrus, once you get the feel of it check the books that math departments use for teaching wavelets.

I bought this book with a great enthusiasm and hope to learn something about filter banks. Soon after I started to read it I figured that the book has a great deal of inconsistencies. To begin with the notation is awkward, and the dimensions of the vectors and matrices are often infinite! Clearly implying that the authors are not very practical after all... Chapter 4 on filter banks is more or less OK, with mentioning of many types and choices for filter designs but nothing deep and concrete (In Page 121 I couldn't find the reference "[VK]"). Chapter 5 is a disaster: Title of Section 5.2 should be "Orthogonal" instead of "Orthonormal", Page 148 says "presented below" but there are nothing "below", other mathematical proofs are sloppy and invalid. I recommend people interested in filter banks to read Vaidyanathan's book instead.

Ok. I am divulging a secret here. This book is a hidden Gem, who never actually realized its potential, it seems. It packs so much crucial / critical information about DSP, multirate techniques, filter design, multirate system optimization that its simply incredible !!! I however agree with the reviews which rate its difficulty high, and a reviewer claiming that it has been written in bouts of inspiration. I can associate with him, and yet, associate more with the authors, when they set about writing on this topic. Its a huge undertaking for one. The literature was/is exploding all the time. To give a composite picture, they are freaking hard pressed. No wonder, people find it tiring, and give up before they could realize what a gem of a creation it is. Gilbert strang, is really at his best here. He is plunging left , right and center, grasping concepts in his solid hands, and massaging them for palatability. I found the book intimidating too, at first, but looked at other texts. The gaps that others would always leave, this book would come in and fill them, not only fill them, but solidify them, re-enforce them with pure concrete.
Since I said I will be divulging , here is my advice on people starting out with it. Dont take the progression of book too seriously. Its not that its flawed, its just that one cannot appreciate its beauty at the outset. So , do not cover the material as it presents itself. Atleast not the whole of it. And there can be different points of emphasis for people approaching from different directions, but generally, try to glean important DSP details, and leave the multiresolution stuff for a later date. One should keep coming back to it every now and then, and you would see it makes more sense. Chapter three, four and five. And most expecially FOUR. Thats the meat of the book, and truly striking. But to get struck, you have to be listening closely to him in Chapter 2. Chapter 3 is a meticulous, nowhere ever to be found treatment of sampling with z operators. Pay very close attention. Once you hit Chapter 4 then, you will be able to appreciate the perfect reconstruction framework. Chapter 4 then trails off, and emphasized alias cancellation a little too much. Again no fault of authors, just the way things stand in literature. Kind of misses the perfect reconstruction condition, and finishes off introducing polyphase notation. You have to keep in mind that perfect reconstruction conditions still need to be satisfied and hence all the later chapters. Keep coming back to three and two whenever you feel you are getting lost.
I suggest Vitterli's subband coding and Fred Harris multirate communications are very helpful texts to go with it, for people implementing with a practical flavor in mind. Esp Fred Harris gives a totally striking transciever view of the whole theory. Can help tie things together more beautifully.
In short, the reviews dont quite do justice to this momentous text. Its trully striking. Its truly a work of love. And absolutely dead instructive. Instruction doesnt get any better.

The book start from very basic concept, and then step by step lead you to wavelet, and end at applications.
You need only undergraduate level DSP knowledge to learn this.

I have taken several courses on DSP. I found that this text takes topics that were presented, expanded them in great detail. It took some of the mystery out of my previous presentations. Anyone working with DSP should have this on their shelf!

After reading several other books on wavelets, I picked up this one. On first inspection, it appears to be reader friendly and practical. It is neither! The presentation is disjointed, unclear, and lacks depth. I think it would be very hard to learn about wavelets from this book.